English

Improving Code-switching Language Modeling with Artificially Generated Texts using Cycle-consistent Adversarial Networks

Computation and Language 2021-12-14 v1

Abstract

This paper presents our latest effort on improving Code-switching language models that suffer from data scarcity. We investigate methods to augment Code-switching training text data by artificially generating them. Concretely, we propose a cycle-consistent adversarial networks based framework to transfer monolingual text into Code-switching text, considering Code-switching as a speaking style. Our experimental results on the SEAME corpus show that utilising artificially generated Code-switching text data improves consistently the language model as well as the automatic speech recognition performance.

Keywords

Cite

@article{arxiv.2112.06327,
  title  = {Improving Code-switching Language Modeling with Artificially Generated Texts using Cycle-consistent Adversarial Networks},
  author = {Chia-Yu Li and Ngoc Thang Vu},
  journal= {arXiv preprint arXiv:2112.06327},
  year   = {2021}
}

Comments

4 pages, 1 figure, Interspeech 2020

R2 v1 2026-06-24T08:14:10.114Z